Technically, he kept chasing improvements. Optimized elliptic-curve arithmetic, memory-efficient key representations, better heuristics to eliminate impossible candidates. He mapped the search space in diagrams and probability charts: expected collisions, false-positive rates, the math that made success almost impossible except at the edges of human error. He calculated the cost — electricity, hardware, time — and found that even with cutting-edge ASICs and clusters, the chance of stumbling on a randomly chosen private key remained astronomically small. The honest conclusion wasn’t thrilling: for properly-random keys, brute force is fantasy. The meaningful targets were leaks, mistakes, and the small seams in human systems.